You've Got RFID ... Now What?
Author:
January 2007
The surge of the RFID technology during the past year has created growth in the number of different organizations manufacturing RFID readers, tags and accessories. While this provides welcome competition in the market and keeps the prices down, it also creates a dilemma in organizations who must decide which product to choose and how to integrate the products in their current business scenarios. The main problem in the implementation of RFID in current business situations is the development of applications around an array of devices that combined well together to implement the business logic. In order to achieve this, custom applications need to be developed that use the data obtained from the sensors and integrate them in real time to make business decisions. This process, however, requires thorough understanding of the data generated by the sensor and as well as the hardware and software interfaces of the devices. The current strategy for dealing with this added layer of complexity is to purchase and deploy additional software technically known as "edgeware" or "middleware". Unfortunately, the acquisition of such software does not solve the problem but actually adds a new layer to the problem itself. Organizations often need to spend additional resources on acquiring training to be able to use the middleware.
The RFID Lab at the Raj Soin College of Business at Wright State University is developing a solution to this problem which might be ideal for small to medium sized organizations. This project addresses the "middleware" problem from two perspectives: first, it will provide a model for capturing the control logic behind RFID-driven applications, and second, it will provide an architecture that allows applications to be created based on this model without the need for a considerable amount of development efforts. The significance of this project comes from the fact that the resources saved in the development of applications will result in faster development cycles and higher dependability of applications. Thi will lead to faster realization of the return on investment and the higher business process automation that RFID technology promises to deliver.
FlexRFID, the system that is under development, is a combination of multiple related underlying concepts. FlexRFID itself is a control logic model, which can be used to capture the control aspects of business systems. FlexRFID depends on an underlying data architecture RTTA (Resource Time Tracking Architecture) which captures the data obtained from event processing on the sensor network. This data is then analyzed using advanced mining techniques and query languages for ad-hoc extraction of information.
Acquiring RFID interrogators and transponders is only the first stage of developing a complete solution. A significant effort is expended toward setting up the system logic using the different types of sensors to ensure that the business rules are properly implemented. This is typically performed by developing applications or hiring system integrators to implement the appropriate logic in proprietary and often poorly designed and understood code. FlexRFID includes a conceptual model for capturing complex control logic using a small set of graphical control patterns. These patterns can be arranged in a control Finite State Machine (FSM) diagram that describes the behavior of the system. Theoretically, this type of a FSM is Finite Automaton with output. Two types of such machines are known in theoretical computation literature: (i) Moore machines or Sequential Machines where the outputs are associated with states, and (ii) Mealy machines, where outputs are associated with transitions, but these can be shown to be equivalent [Moore, 1956]. The current proof-of-concept uses Moore machines to capture the control model of any RFID and Sensor controlled system. The FSMs can then be executed to quickly test and troubleshoot applications, with actual hardware or simulated devices. The FSM control is system and vendor independent, so can be mapped to most RFID and other sensor manufacturers. Since the models have a simple graphical representation and a small underlying binary representation, they are easy to understand for users, as well as simple to embed into RFID readers for automated processing. Figure 1-a shows a FlexRFID model, 1-b shows a rendering of the model and execution on simulated devices using the FlexRFID software, and 1-c shows how FlexRFID can be used to drive real hardware devices.
To find out more information about this project, please contact Dr. Arijit Sengupta at This e-mail address is being protected from spambots. You need JavaScript enabled to view it .
